With 512K-GPU clusters becoming the norm for AI factories, copper is hitting its limits. An OFC 2026 panel showed how CPO delivers a 3.5X efficiency gain, and how the 12.8T XPO rewrites the switch-density game.

Google's fungible data center lets one facility host any generation of TPU, GPU, or CPU. As OCP standardizes power, cooling, and facility specs, Taiwan's thermal and power suppliers face a new spec-in battle, and the ticket in is OCP: a higher bar than capacity.

Pluggable optics are hitting the front-panel limit, not a speed limit. Andy Bechtolsheim sandwiches a cold plate between two paddle cards to shrink 102.4T from 4U to 1U. His only doubt about CPO is attach yield, not the physics, and a Foxconn connector now sampling decides whether XPO works at 400G/lane.

At OFC 2026, Amphenol showed its vision for AI data center interconnect: the XPO module packs 8x the density of OSFP with built-in liquid cooling, challenging CPO's lead as the industry moves from 224G to 448G.